Graphics Cards

[sur6e]

Hi,

I know that 110C is considered the safe max for hotspot/junction temps but lately I've seen my card hit 113C in HWinfo64 and I have also had a few random lockups in the last few days. Sound like discord keeps working for a while in the background but everything on screen freezes, mouse doesn't move, ctrl alt del, alt tab, etc does nothing. I have to press the reset button to reboot. I don't know if the two are related but I was stable and having no issues. My main question is if this is a warranty eligible issue and what I should expect from Gigabyte support (if anyone has experience with that). My warranty goes till late 2024 I believe. The card has always been on the hot side and I have always seen 110C max, but now it's going over a little. I used to undervolt an crank up the fans a bit but with all the driver/MPO issues in recent months I don't really know what's stable anymore so I've been trying to run it at default settings, maybe just higher fans. Thanks.

[elstaci]

If you HAVE GOOD AIR CIRCULATION inside your PC then I suggest you open a Gigabyte Warranty to see if the Thermal Paste needs to be replaced on the GPU card.

Many Users are able to do this themselves BUT since you still have a Warranty on the GPU card it is best to go through your Gigabyte Warranty to have it done.

Your Warranty might be voided if you do it yourself.

Also make sure your GPU card's Fans are clean and turns easily without noise manually with Power off to the PC. 

Talking about Fans see if you can create a more aggressive Fan Curve in Radeon Settings that makes the Fan go to full speed at a lower Temperature to see if that help or not.

NOTE: Your GPU card is throttling at 113c and due to overheating past 113c the GPU card is probably causing the freezes and crashes on your PC.

[sur6e]

Thanks. Yeah, I have two 140mm fans taking in air directly onto the GPU fans from the bottom, and three 120mm fans taking in air from the side. Then I have a 280mm radiator in the top for the CPU with its fans serving as exhaust. I think airflow is good enough. I use Fan Control to ramp up the bottom fans when the GPU temp goes up and the side fans when the VRM's warm up, and the exhaust ramps up when the CPU temp goes up. The default fan setting is 42% in Radeon software. I don't want the thing screaming at me but have ramped it up a good bit before. I'll test again though. I indeed didn't want to try repasting/padding because it's not clear to me if that voids warranty. I'll test some more and contact support.

[elstaci]

I would first open a Gigabyte Warranty Ticket to see if your GPU card needs to be RMAed.

Not sure about the RX5700XT but on some GPU cards if you remove the Cowl (Heatsink) to re-paste you also will need to replace the Thermal pads.

NOTE: If it was working normally below 110c and then started to overheat probably does indicate the Thermal paste is not working efficiently anymore.  Many times the Manufacturers doesn't really apply a high quality expensive thermal paste on their products to save money, but that is just my own opinion.

Most likely your Warranty would be voided if you try to re-paste the GPU chip yourself. But I would ask though just in case it wouldn't.

[elstaci]

Also check your GPU Card's Heat sink fins to make sure they aren't full of dust or dirt or has a oily film on them (smoke from cooking (-: ).

[sur6e]

Just got an auto-reply saying everything is closed till 1/3. Guess I won't see a reply till then.

[sur6e]

I tested in BF2042 and had the GPU fans at 100% (over 4k rpm) and junction temps still reached at least 105C from what I saw. Didn't have HWinfo open, only Radeon software.